WIP: Implement the 'gradient' method of ElasticEnergyFunctional

Sander, Oliver requested to merge implement-gradient into master

And use it in ElasticEnergyConstrainedLinearization.

Unlike the dune-functions style, we do not return a function object for the gradient. We simply compute the gradient at the given point, and return the vector containing its values. This saves us one level of indirection, and ElasticEnergyConstrainedLinearization does not need more.

Merge request reports